Abstract: The thermal infrared band data of Landsat6 TM/ETM⁺ images acquired in 1995, 2002 and 2010 were used to invert the land surface temperature of Hefei. The values of the land surface temperature were normalized, and the grading figure of the land surface temperature was obtained through the standard deviation taxonomy. Furthermore, the NDVI, landuse/landcover change, and social economic development data were used to analyze the spatiotemporal variation and its causes of heat island effect in Hefei. Overall, the urban heat island effect in the city presented a spatial pattern of being higher in northeast-southwest and lower in southeast-northwest. The areas with strong heat island effect were mainly concentrated in the old city, and extend to the eastern and southern new industrial areas, showing a new characteristic of multi-center and patchy distribution. In 1995-2010, the spatial distribution and extension of the heat islands were basically in consonance with the layout of the urban districts in Hefei. The scope of the heat islands was expanding, but the proportions of high temperature area and extremely high temperature area were declined. The intensity of the urban heat islands exhibited a decreasing trend. There existed an obvious negative correlation between the LST and NDVI. The construction land had the highest average land surface temperature, followed by the farmland and woodland, and by the water body. The construction land had the priority position in the highest three grades of land surface temperature, which made a great contribution to the heat island effect. The water body and woodland played the most stable role in reducing the urban surface temperature.